CareCompanion AI

Making healthcare understandable for everyone, everywhere. Your personal AI health assistant. Upload lab reports or prescriptions and get instant, plain-English explanations powered by Gemini 3.

Comment

Inspiration

Reading medical documents feels like learning a new language. Lab reports with "Hematocrit," "Basophils," and "Creatinine" are confusing and scary for patients. We wanted to build a bridge between complex medical data and patient understanding—a companion that not only reads data but also explains it with empathy and clarity.

What it does

CareCompanion AI is a multimodal health assistant that:

  1. Analyzes Medical Documents: Upload images or PDFs of lab reports (blood tests, metabolic panels, etc.).
  2. Explains in Plain English: Uses Gemini 3's deep reasoning to break down complex medical terms into simple analogies (e.g., "White blood cells are your body's defense team").
  3. Voice Interaction: Patients can ask follow-up questions naturally using their voice ("Is my cholesterol strictly bad?").
  4. Fact-Checked Answers: Leverages Google Search Grounding to verify facts and check drug interactions against the latest medical guidelines.

How we built it

  • Frontend: Next.js 14 (App Router) with Tailwind CSS and shadcn/ui for a modern, accessible interface.
  • AI Engine: Gemini 3 Flash Preview via the Google AI Node.js SDK.
  • Multimodal Capabilities: Used Gemini's vision capabilities to process document images directly (no OCR needed!).
  • Grounding: Integrated Google Search tool to ensure answers are factually grounded.
  • Voice: Web Speech API for real-time voice input.

Challenges we ran into

  • Prompt Engineering: Tuning the model to be "empathetic but precise" was tricky—we didn't want it to sound like a robot or give false hope.
  • Rate Limits: Working with the preview model meant hitting quota limits often, forcing us to optimize our context usage!

Accomplishments that we're proud of

  • Seamless drag-and-drop document analysis that feels like magic.
  • The "Deep Think" style explanations that use analogies (e.g., Hemoglobin = Oxygen Trucks).
  • Integrating Voice + Vision + Search in a single seamless flow.

What's next for CareCompanion AI

  • Multilingual Support: Explaining reports in parents' native languages.
  • History Tracking: improved long-term memory of patient history.
  • EHR Integration: Connecting directly to hospital systems.

Built With

Share this project:

Updates